* KASAN: use-after-free Read in io_rsrc_node_ref_zero
@ 2021-12-27 16:47 Palash Oswal
0 siblings, 0 replies; only message in thread
From: Palash Oswal @ 2021-12-27 16:47 UTC (permalink / raw)
To: io-uring, LKML, syzkaller-bugs, Pavel Begunkov, Dmitry Vyukov,
Hillf Danton, Jens Axboe
Hello,
My instance of syzkaller found the following issue on v5.15.0 tree.
Git Tree : stable
Unfortunately I have not been able to create a consistent C reproducer
for this issue yet. I could use some help in simplifying the syz-repro
here. This bug hasn't been identified by the syzbot instance yet.
BUG: KASAN: use-after-free in io_rsrc_node_ref_zero+0x5b/0x2f0
fs/io_uring.c:7656
Read of size 8 at addr ffff888001d3a780 by task swapper/0/0
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.15.0 #13
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
Call Trace:
<IRQ>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x5d/0x80 lib/dump_stack.c:106
print_address_description+0x88/0x3e0 mm/kasan/report.c:256
__kasan_report mm/kasan/report.c:442 [inline]
kasan_report+0x1a2/0x1f0 mm/kasan/report.c:459
__asan_report_load8_noabort+0x14/0x20 mm/kasan/report_generic.c:309
io_rsrc_node_ref_zero+0x5b/0x2f0 fs/io_uring.c:7656
percpu_ref_put_many include/linux/percpu-refcount.h:322 [inline]
percpu_ref_put include/linux/percpu-refcount.h:338 [inline]
percpu_ref_call_confirm_rcu lib/percpu-refcount.c:163 [inline]
percpu_ref_switch_to_atomic_rcu+0x56f/0x5e0 lib/percpu-refcount.c:205
rcu_do_batch kernel/rcu/tree.c:2508 [inline]
rcu_core+0xf58/0x1500 kernel/rcu/tree.c:2743
rcu_core_si+0xe/0x10 kernel/rcu/tree.c:2756
__do_softirq+0x216/0x516 kernel/softirq.c:558
invoke_softirq kernel/softirq.c:432 [inline]
__irq_exit_rcu+0x88/0x120 kernel/softirq.c:636
irq_exit_rcu+0xe/0x10 kernel/softirq.c:648
sysvec_apic_timer_interrupt+0x76/0x90 arch/x86/kernel/apic/apic.c:1097
</IRQ>
asm_sysvec_apic_timer_interrupt+0x12/0x20
RIP: 0010:native_safe_halt+0xf/0x20 arch/x86/include/asm/irqflags.h:52
Code: e1 07 80 c1 03 38 c1 7c ba 48 89 df e8 9a 01 fd fd eb b0 00 00
cc cc 00 00 cc cc 55 48 89 e5 eb 07 0f 00 2d 53 fe 49 00 fb f4 <5d> c3
66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 e5 eb
RSP: 0018:ffffffffa3007d68 EFLAGS: 00000206
RAX: 000000000007c07e RBX: 0000000000000000 RCX: 000000000007c07e
RDX: 0000000000000001 RSI: 0000000000000004 RDI: 0000000000000001
RBP: ffffffffa3007d68 R08: dffffc0000000000 R09: ffffed100450631d
R10: ffffed100450631d R11: ffffffffa1fd6960 R12: 0000000000000000
R13: ffffffffa301bd40 R14: 1ffffffff4600fc4 R15: dffffc0000000000
arch_safe_halt arch/x86/kernel/process.c:715 [inline]
default_idle+0xe/0x20 arch/x86/kernel/process.c:716
arch_cpu_idle+0x13/0x20 arch/x86/kernel/process.c:708
default_idle_call+0x79/0x1c0 kernel/sched/idle.c:112
cpuidle_idle_call kernel/sched/idle.c:194 [inline]
do_idle+0x1f0/0x5c0 kernel/sched/idle.c:306
cpu_startup_entry+0x25/0x30 kernel/sched/idle.c:403
rest_init+0x10c/0x130 init/main.c:734
arch_call_rest_init+0x13/0x15
start_kernel+0x3f2/0x468 init/main.c:1142
x86_64_start_reservations+0x24/0x26 arch/x86/kernel/head64.c:525
x86_64_start_kernel+0x7c/0x7f arch/x86/kernel/head64.c:506
secondary_startup_64_no_verify+0xb1/0xbb
Allocated by task 1639:
kasan_save_stack mm/kasan/common.c:38 [inline]
kasan_set_track mm/kasan/common.c:46 [inline]
set_alloc_info mm/kasan/common.c:434 [inline]
____kasan_kmalloc+0xc4/0xf0 mm/kasan/common.c:513
__kasan_kmalloc+0x9/0x10 mm/kasan/common.c:522
kasan_kmalloc include/linux/kasan.h:264 [inline]
kmem_cache_alloc_trace+0x201/0x2d0 mm/slub.c:3240
kmalloc include/linux/slab.h:591 [inline]
kzalloc include/linux/slab.h:721 [inline]
io_rsrc_data_alloc+0x68/0x520 fs/io_uring.c:7794
io_sqe_files_register+0x21f/0xf30 fs/io_uring.c:8231
__do_sys_io_uring_register+0xb84/0x3460 fs/io_uring.c:10945
__se_sys_io_uring_register fs/io_uring.c:10925 [inline]
__x64_sys_io_uring_register+0x9f/0xb0 fs/io_uring.c:10925
do_syscall_64+0x48/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xae
Freed by task 1640:
kasan_save_stack mm/kasan/common.c:38 [inline]
kasan_set_track+0x4c/0x80 mm/kasan/common.c:46
kasan_set_free_info+0x23/0x40 mm/kasan/generic.c:360
____kasan_slab_free+0x110/0x150 mm/kasan/common.c:366
__kasan_slab_free+0x11/0x20 mm/kasan/common.c:374
kasan_slab_free include/linux/kasan.h:230 [inline]
slab_free_hook mm/slub.c:1700 [inline]
slab_free_freelist_hook+0x98/0x160 mm/slub.c:1726
slab_free mm/slub.c:3492 [inline]
kfree+0xda/0x290 mm/slub.c:4552
io_rsrc_data_free fs/io_uring.c:7783 [inline]
__io_sqe_files_unregister fs/io_uring.c:7861 [inline]
io_sqe_files_unregister fs/io_uring.c:7874 [inline]
__io_uring_register fs/io_uring.c:10829 [inline]
__do_sys_io_uring_register+0x2172/0x3460 fs/io_uring.c:10945
__se_sys_io_uring_register fs/io_uring.c:10925 [inline]
__x64_sys_io_uring_register+0x9f/0xb0 fs/io_uring.c:10925
do_syscall_64+0x48/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xae
Last potentially related work creation:
kasan_save_stack+0x36/0x60 mm/kasan/common.c:38
kasan_record_aux_stack+0xbc/0xe0 mm/kasan/generic.c:348
insert_work+0x58/0x2d0 kernel/workqueue.c:1353
__queue_work+0x911/0xc90 kernel/workqueue.c:1519
queue_work_on+0x5d/0xa0 kernel/workqueue.c:1546
queue_work include/linux/workqueue.h:501 [inline]
call_usermodehelper_exec+0x240/0x3e0 kernel/umh.c:435
call_modprobe kernel/kmod.c:98 [inline]
__request_module+0x34b/0x790 kernel/kmod.c:170
dev_load+0x5d/0xb0 net/core/dev_ioctl.c:446
dev_ioctl+0x528/0x10d0 net/core/dev_ioctl.c:511
sock_do_ioctl+0x2ee/0x510 net/socket.c:1132
sock_ioctl+0x56d/0x950 net/socket.c:1235
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:874 [inline]
__se_sys_ioctl+0xed/0x150 fs/ioctl.c:860
__x64_sys_ioctl+0x80/0x90 fs/ioctl.c:860
do_syscall_64+0x48/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xae
Second to last potentially related work creation:
kasan_save_stack+0x36/0x60 mm/kasan/common.c:38
kasan_record_aux_stack+0xbc/0xe0 mm/kasan/generic.c:348
insert_work+0x58/0x2d0 kernel/workqueue.c:1353
__queue_work+0x911/0xc90 kernel/workqueue.c:1519
queue_work_on+0x5d/0xa0 kernel/workqueue.c:1546
queue_work include/linux/workqueue.h:501 [inline]
call_usermodehelper_exec+0x240/0x3e0 kernel/umh.c:435
call_modprobe kernel/kmod.c:98 [inline]
__request_module+0x34b/0x790 kernel/kmod.c:170
dev_load+0x96/0xb0 net/core/dev_ioctl.c:448
dev_ioctl+0x528/0x10d0 net/core/dev_ioctl.c:511
sock_do_ioctl+0x2ee/0x510 net/socket.c:1132
sock_ioctl+0x56d/0x950 net/socket.c:1235
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:874 [inline]
__se_sys_ioctl+0xed/0x150 fs/ioctl.c:860
__x64_sys_ioctl+0x80/0x90 fs/ioctl.c:860
do_syscall_64+0x48/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xae
The buggy address belongs to the object at ffff888001d3a780
which belongs to the cache kmalloc-96 of size 96
The buggy address is located 0 bytes inside of
96-byte region [ffff888001d3a780, ffff888001d3a7e0)
The buggy address belongs to the page:
page:000000008e5b5d32 refcount:1 mapcount:0 mapping:0000000000000000
index:0x0 pfn:0x1d3a
flags: 0xfffffc0000200(slab|node=0|zone=1|lastcpupid=0x1fffff)
raw: 000fffffc0000200 ffffea00000d4100 0000000c0000000c ffff888001041780
raw: 0000000000000000 0000000000200020 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff888001d3a680: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
ffff888001d3a700: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
>ffff888001d3a780: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
^
ffff888001d3a800: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
ffff888001d3a880: fa fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
==================================================================
------------[ cut here ]------------
WARNING: CPU: 0 PID: 81 at fs/io_uring.c:9283 io_ring_ctx_free
fs/io_uring.c:9283 [inline]
WARNING: CPU: 0 PID: 81 at fs/io_uring.c:9283
io_ring_exit_work+0x1cac/0x1e60 fs/io_uring.c:9446
Modules linked in:
CPU: 0 PID: 81 Comm: kworker/u2:1 Tainted: G B 5.15.0 #13
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
Workqueue: events_unbound io_ring_exit_work
RIP: 0010:io_ring_ctx_free fs/io_uring.c:9283 [inline]
RIP: 0010:io_ring_exit_work+0x1cac/0x1e60 fs/io_uring.c:9446
Code: 48 8b 04 25 28 00 00 00 48 3b 84 24 20 01 00 00 0f 85 b4 01 00
00 48 8d 65 d8 5b 41 5c 41 5d 41 5e 41 5f 5d c3 e8 e4 84 92 ff <0f> 0b
e9 38 f7 ff ff e8 d8 84 92 ff 0f 0b e9 5d f7 ff ff e8 cc 84
RSP: 0018:ffff888002217b80 EFLAGS: 00010293
RAX: ffffffffa01c677c RBX: ffff88800191e558 RCX: ffff888002209a00
RDX: 0000000000000000 RSI: 0000000000000008 RDI: ffff888022830340
RBP: ffff888002217d00 R08: dffffc0000000000 R09: ffffed1000323ca0
R10: ffffed1000323ca0 R11: 0000000000000000 R12: ffff888004935f28
R13: ffff88800191e4e8 R14: 0000000000000000 R15: dffffc0000000000
FS: 0000000000000000(0000) GS:ffff888022800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f4ef266c000 CR3: 00000000048f6001 CR4: 0000000000370ef0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
process_one_work+0x6c8/0x1050 kernel/workqueue.c:2297
worker_thread+0x9f1/0x1520 kernel/workqueue.c:2444
kthread+0x3af/0x4a0 kernel/kthread.c:319
ret_from_fork+0x1f/0x30
---[ end trace be1e878fc6a839ea ]---
----------------
Code disassembly (best guess):
0: e1 07 loope 0x9
2: 80 c1 03 add $0x3,%cl
5: 38 c1 cmp %al,%cl
7: 7c ba jl 0xffffffc3
9: 48 89 df mov %rbx,%rdi
c: e8 9a 01 fd fd callq 0xfdfd01ab
11: eb b0 jmp 0xffffffc3
13: 00 00 add %al,(%rax)
15: cc int3
16: cc int3
17: 00 00 add %al,(%rax)
19: cc int3
1a: cc int3
1b: 55 push %rbp
1c: 48 89 e5 mov %rsp,%rbp
1f: eb 07 jmp 0x28
21: 0f 00 2d 53 fe 49 00 verw 0x49fe53(%rip) # 0x49fe7b
28: fb sti
29: f4 hlt
* 2a: 5d pop %rbp <-- trapping instruction
2b: c3 retq
2c: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1)
33: 00 00 00
36: 0f 1f 44 00 00 nopl 0x0(%rax,%rax,1)
3b: 55 push %rbp
3c: 48 89 e5 mov %rsp,%rbp
3f: eb .byte 0xeb
I think the bug might be the call of kfree(ref_node) in
https://elixir.bootlin.com/linux/v5.15/source/fs/io_uring.c#L7686
while a concurrent thread uses the structure.
Syz repro:
# {Threaded:true Collide:true Repeat:true RepeatTimes:0 Procs:1
Slowdown:1 Sandbox:none Leak:false NetInjection:true NetDevices:true
NetReset:true Cgroups:true BinfmtMisc:true CloseFDs:true KCSAN:false
DevlinkPCI:false USB:false VhciInjection:false Wifi:false
IEEE802154:false Sysctl:true UseTmpDir:true HandleSegv:true
Repro:false Trace:false LegacyOptions:{Fault:false FaultCall:0
FaultNth:0}}
r0 = io_uring_setup(0x70c1, &(0x7f0000000000))
io_uring_register$IORING_REGISTER_FILES(r0, 0x2, &(0x7f0000000880), 0x21)
io_uring_register$IORING_REGISTER_FILES_UPDATE(r0, 0x6, 0x0, 0x0)
r1 = open$dir(&(0x7f0000000000)='./file0\x00', 0x23c41, 0x0)
write$binfmt_script(r1, 0x0, 0x2a)
open(&(0x7f00000001c0)='./file0/file0\x00', 0x800, 0x142)
io_uring_register$IORING_UNREGISTER_FILES(r0, 0x3, 0x0, 0x0)
io_uring_register$IORING_REGISTER_BUFFERS(r0, 0x0,
&(0x7f0000000380)=[{0x0}], 0x1)
r2 = io_uring_setup(0x7628, &(0x7f0000000080)={0x0, 0x7628})
r3 = open$dir(&(0x7f0000000000)='./file0\x00', 0x8840, 0x0)
io_uring_register$IORING_REGISTER_FILES(r2, 0x2,
&(0x7f0000000100)=[r3, 0xffffffffffffffff], 0x2)
r4 = io_uring_setup(0x7628, &(0x7f0000000080)={0x0, 0x7628})
r5 = open$dir(&(0x7f0000000000)='./file0\x00', 0x8840, 0x0)
io_uring_register$IORING_REGISTER_FILES(r4, 0x2,
&(0x7f0000000100)=[r5, 0xffffffffffffffff], 0x2)
r6 = open$dir(&(0x7f0000000080)='./file0/file0\x00', 0x28000, 0x104)
io_uring_register$IORING_REGISTER_FILES_UPDATE(r0, 0x6,
&(0x7f0000000180)={0x0, 0x0, &(0x7f0000000140)=[r3,
0xffffffffffffffff, r1, r4, r6]}, 0x5)
io_uring_register$IORING_UNREGISTER_BUFFERS(r0, 0x1, 0x0, 0x0)
Kernel Build Config:
https://gist.github.com/oswalpalash/18e847d6e24e3452bc811526fd6f76bb
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-12-27 21:48 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2021-12-27 16:47 KASAN: use-after-free Read in io_rsrc_node_ref_zero Palash Oswal
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox